The Underwriting Shortcut Smart Investors Use

Stacking Trades — Mon, 12 Jan 2026 20:34:06 +0000

Walk into almost any deal room and you’ll hear the same opening lines. The market is taking off. The product feels inevitable. The team has momentum. It’s familiar language because it does the job: it grabs attention, pulls in capital, and buys the company a little more runway.

But the market eventually asks a different question, one that has nothing to do with the story and everything to do with the mechanism: what actually creates value here. Not in theory. In the numbers, over time, under pressure.

You can watch this shift in real time in public companies. When Apple filed its 2025 annual report, it disclosed a gross margin percentage of 46.9% (up from 46.2% the prior year), and described drivers like favorable costs and mix. That is what an engine looks like from the outside: a business that can hold or expand margin at scale.

The mistake investors make is treating returns as a surprise. In most durable outcomes, the return was being built quietly by a small set of repeatable levers. Your job is to find which lever is doing the work.

The business is the engine, not the story

A company can feel unstoppable for a stretch just on story, especially when investors are paying for what it could become. But that optimism eventually has to turn into something real and measurable: the ability to hold price, make money on each customer, keep customers around and grow them over time, or acquire new customers efficiently without the costs exploding.

Those are not “framework words.” They are the few ways businesses consistently turn effort into cash flow, and cash flow into enterprise value. Everything else is supporting cast.

Pricing power is the cleanest form of leverage

Pricing power is when a company can charge more, or nudge customers toward higher-priced options, and people still keep buying. You see it when churn doesn’t spike after a price increase, conversion holds steady, and margins quietly get better because the business can defend what it charges.

Streaming is a clean example of how pricing power looks in practice. In the UK, Netflix raised subscription prices in early 2025, with reporting noting the increase and the company’s rationale of reinvesting in content and product. Whether you like the product or not, the financial point is clear: companies that can lift prices and keep a large customer base engaged have a fundamentally different earnings profile than companies that must discount to grow.

This is also why margin disclosures matter. Apple’s reported gross margin level and commentary around mix and costs are not just accounting lines. They are the visible output of a pricing and product strategy that investors can measure over time.

Unit economics decide whether growth is building value

Revenue growth can be bought. Value creation cannot.

Unit economics are about what is left after the direct costs required to deliver the product or service. When unit economics are improving, growth tends to become less fragile. When unit economics are deteriorating, growth often becomes a financing strategy: the company needs new capital to keep the machine running.

A good sign unit economics are getting healthier is when a company finds ways to serve customers more cheaply without making the experience worse. That’s why the most meaningful “efficiency” wins usually show up in the day-to-day operations, not in the marketing. Klarna, for instance, said its AI assistant handled two-thirds of customer service chats in its first month and cut resolution time compared with how those requests were handled before. Whether or not you buy every claim, it’s a real example of the basic point: change the workflow, and the cost to serve can shift fast.

Retention and expansion turn customers into an asset

A business with weak retention has to keep running just to stay in place. A business with strong retention and expansion can grow even if new customer acquisition gets harder.

You see this whenever a company gets better at earning money from people who are already using the product. Netflix’s password-sharing crackdown gets talked about as a growth move, but the investor takeaway is really about retention economics: turning existing usage into paid accounts without killing engagement in the process.

In underwriting terms, retention is what transforms marketing spend from a cost into a long-lived asset. Expansion is what allows the business to improve even when top-line growth slows. Cohorts are where this becomes obvious, quickly, if you can get them.

“If you can’t name the engine, you’re not underwriting. You’re hoping.”

Distribution efficiency is where most stories collapse

Distribution is not “marketing.” It is the system that turns attention into revenue at a cost the company can sustain.

When distribution is strong, you see stable or improving payback periods as spend scales, and you see growth that does not require perpetual promotional pressure. When distribution is weak, CAC rises, payback stretches, and the business quietly becomes dependent on perfect market conditions.

Operationally, the best companies reduce friction across the chain that connects demand to delivery. That includes customer support, where delays and reroutes create churn you will never see in a glossy deck. Lyft said its Claude-powered customer care assistant reduced average resolution time by 87%. If those kinds of gains hold, they do not just cut costs. They protect conversion and retention, which is distribution by another name.

The one-chart test: the engine should be visible

A return engine should be visible in a small number of views.

If the engine is pricing power, it should show up in margins that hold up and price increases that don’t scare customers away. If it’s unit economics, you should see more profit left over per customer and fewer dollars bleeding out through support, fulfillment, or churn. If it’s retention and expansion, cohorts should stick around and, ideally, spend more over time. And if the edge is delivery speed, you can usually get a good read from operational measures like the DORA metrics: how often teams ship, how long changes take, how often releases break things, and how quickly they recover.

This is not about oversimplifying. It is about refusing to let a story substitute for a mechanism.

What breaks is usually predictable

The failure modes repeat: growth that is mostly discounting, “partnerships” that never show up in conversion, profitability that exists only because costs were deferred, and KPIs that improve only when definitions change.

A better habit is to keep asking, from the start, what has to stay true for this business to keep working. Then focus on a small handful of variables that actually tell you whether that’s still happening. When the engine is real, you’ll see it show up in the filings, in how customers behave, and in the company’s day-to-day performance. When it isn’t, the pitch usually just gets louder to make up for it.

Trading FLOPs

Stacking Trades — Fri, 07 Nov 2025 22:45:38 +0000

The Invisible Commodity

Training the latest generative-AI models costs billions, not millions. By mid-2025, leading labs reported that compute budgets had surged, driven by rising demand for high-performance GPUs and data-center scale-up. For example, CoreWeave signed a five-year contract worth US $11.9 billion with OpenAI in March 2025.

In September, CoreWeave strengthened its partnership with OpenAI for up to $6.5 billion. This brings their total deal to about $22.4 billion.

In essence: compute has become a bottleneck, and bottlenecks trade like commodities.

From Cloud to Market

What started off as dealing with cloud compute as a cost center is now turning into a way of looking at it as a tradable asset. Nvidia announced the new software platform Lepton, which will create a marketplace for cloud-based AI chip capacity, thus marking the formalization of compute-markets.

Meanwhile, the global “GPU as a Service” market is projected to grow from US $6.54 billion in 2024 to US $8.21 billion in 2025, and further to US $26.62 billion by 2030 (CAGR ~26.5%).

Running a training job is no longer just a tech operation—it’s a market signal.

“We used to buy capacity; now we’re trading it.”

Liquidity of Intelligence

Investors aren’t just backing software anymore; they’re backing compute infrastructure. Firms treat GPU clusters, training pipelines, and model-hosting farms as utility assets.

The global data-centre GPU market is predicted to grow from about US$119.97 billion in 2025 to US$228.04 billion by 2030.

In other words, compute throughput is becoming the next lever of market advantage.

The Speculation Layer

Compute markets are still thinly regulated but rapidly gaining complexity. For example, Nvidia’s deal to invest up to US $100 billion in OpenAI (and supply chips) shows how compute commitments now resemble infrastructure finance.

The question becomes, if you’re buying access to AI compute, what asset class do you own?

It sits at the intersection of technology, finance, and commodity markets.

The Human Parallel

Commodity trading desks once monitored oil tankers and futures curves. Today some firms are monitoring compute pipelines and GPU lease contracts.

The analogies are real: just as power plants once provided scale for industrial growth, model-labs now seek scale via compute liquidity.

Value is shifting upstream: from software features to underlying infrastructure.

What It Signals

If compute becomes a core asset class:

• Infrastructure providers (not just hardware vendors) gain strategic importance: marketplaces, leasing platforms, and secondary GPU markets become key nodes.

• Software companies face pressure: increasingly the differentiator is access to compute + data, not just features.

• Investors need new metrics: compute utilisation, petaflops per dollar, latency arbitrage become valuable indicators.

• Markets may shift: competition becomes about compute liquidity as much as talent or data.

Investor – Stacking Trades